Several methods exist for quantifying the quality of service provided by a roadway from a bicyclist’s perspective; however, many of these models do not consider physically protected bike lanes and, of those that do, none is based on empirical data from the US. This is problematic as engineers, planners, and elected officials are increasingly looking to objective performance measures to help guide transportation project design and funding prioritization decisions. This paper addresses this gap by presenting a cumulative logistic model to predict user comfort on protected bike lanes developed from data collected during in-person video surveys. The surveys were conducted in Portland, OR with video footage gathered in Chicago, IL, Portland, OR and San Francisco, CA. The model is for road segments only and not signalized intersections. It complements the Highway Capacity Manual 2010 level-of- service methods by providing an analysis procedure for a facility type that is not currently included in the manual. The model indicates that the type of buffer, direction of travel (one-way vs. two-way), adjacent motor vehicle speed limit, and average daily motor vehicle volumes are all significant predictors of bicyclist comfort in protected bike lanes. The model predicts a mean value of ‘A’ or ‘B’ on an A (most comfortable) – F (least comfortable) scale for all protected bike lane clips used in the survey. Consistent with previous research findings, survey respondents report that protected bike lanes are generally more comfortable than other types of on-street infrastructure.